Here's the problem with surveys like this - they are completely voluntary and the only people who will respond are those who may have been bullied. The people who haven't been bullied, don't respond. Thus, your results are skewed NO MATTER WHAT because we don't know how many people didn't respond because they weren't bullied.
Your concern is about a self-selected sample that skews the data because the sample respondents don't truly represent the *entire* population of the target group in Virginia (in this case). You care about the potential for a "random sampling error".
The Margin or Error number that I quoted from the study is the number that tracks and measures how closely the respondent sample tracks to the target population as a whole
. In short, it expresses what the maximum error might be in the results because in statistics no sample (short of 100%) is perfectly representative. This is standard practice. From the Wikipedia
article I linked to to explain why talking about MoE makes sense answering your concern:
The margin of error is a statistic expressing the amount of random sampling error in a survey's results. The larger the margin of error, the less faith one should have that the poll's reported results are close to the "true" figures; that is, the figures for the whole population. Margin of error occurs whenever a population is incompletely sampled.
The study reported the Margin of Error which quantifies exactly how much impact your specific concern has on the study results and I shared "maximum +/- error" scenarios based on their numbers with you:
This means that assuming the maximum margin of error in both directions the 85 percent of gay students who regularly heard homophobic remarks could be as low as 78% or as high as 92%, that the 37 percent who were physically harassed due to their sexual orientation could be as low as 30% and as high as 44%, and that the 21 percent who were physically assaulted could be as low as 14% or as high as 28%. It's most probable that the true number is very close to the actual number reported.
Your primary statistical concern has already been measured and quantified and I've show the maximum impact it can have on the reported results, even though in reality the probability is high that the real number is very close, to one side or the other, to the reported number. The assertion that "the only people who will respond are those who may have been bullied. The people who haven't been bullied, don't respond" is false.
Firstly, GLSEN recruited broadly within LGBT youth communities to achieve a representative sample (from the detailed study report):
GLSEN used two methods to locate survey participants in an effort to obtain a representative national sample of lesbian, gay, bisexual, and transgender (LGBT) youth: outreach through community-based groups serving LGBT youth and outreach via the Internet. With our first method, we randomly selected 50 community-based groups from a list of over 300 groups nationwide and asked their youth participants to complete a paper version of the survey. Our second method utilized GLSENís web presence, e-communications, and online advertising to obtain participants. We posted notices of the survey on LGBT-youth oriented listservs and websites and emailed notices to GLSEN chapters and youth advocacy organizations. To ensure representation of transgender youth, youth of color, and youth in rural communities, we made special efforts to reach out to organizations that serve these populations. We also conducted targeted advertising on the MySpace and Facebook social networking sites. The advertisements targeted users between 13 and 18 years of age who gave some indication on their profile that they were lesbian, gay, bisexual, or transgender.
Secondly, in each of the data points recorded, a percentage of respondents answered in the negative. If your contention had any element of truth, some or all measures of bullying would approach 100%. This is not observed in the results.
There is no statistical foundation whatsoever for asserting that the study as scoped and reported is "meaningless".
After pulling numbers for *all* school children in Virginia instead of the numbers for the LGBT target population of the study in Virginia, you make the following assertion:
So out of around a half a million kids, a self serving study that serves GLSEN's interest, its reason for existence, shows that approximately 0.05% of students have been exposed to homophobic remarks and this is supposed to be newsworthy and generalized across the entire student population?
This study makes no statement about bullying for all school students in Virginia and it collected no data for respondents who were not targets for the study (LGBT kids). It is disingenuous at best to take the percentage of homophobic bullying for lesbian, gay, bi, and trans children and recalculate the percentage based on straight kids who weren't included in the study (and are not the main targets of homophobic bullying) in order to make the problem of anti-gay bullying to LGBT kids seem statistically insignificant.
This kind of distortion and misuse of the data is the definition of bad statistics and bad science. It makes me wonder if your goal is to determine how accurate and true this study is, or to dismiss it and the issue of homophobic bullying wholesale.
If you want to stand by this as the original poster did, that this was indicative of intolerance, be my guest. I would argue the complete and total opposite. This is indicative of extreme tolerance and any homophobia is absolutely outside the norm if such a small percentage of the population is experiencing this.
GLSEN isn't making a case that homophobic bullying affects all children, only that it affects kids who are *actually* lesbian, gay, bi, and transgender (you know, the targets of homophobic bullying) very significantly.
Its not that bad. Its bad for some individuals but for the absolute majority, its not bad at all.
I find your contention here both fallacious and illuminating. By your argument, racist bullying isn't a big problem because it only affects people who are of other races (non-white). The material point is that targeted bullying or any other kind of bias is a significant problem to the target community. Your attempts to say that homophobic bullying is a small problem because only LGBT kids experience it is a total nonsense.
I find organizations like GLSEN specious. I think GLSEN takes individual experiences and extrapolates them out into being the norm and its just not so. This is an example of a poll being used for a press release and THAT'S IT.
This is a methodologically sound study that accurately captures the incidence of homophobic bullying in the LGBT school community and shows how pervasive it is. There is nothing qualitative or anecdotal about this research as you imply. The inability or unwillingness to understand how research is done and the science behind it is not a problem we can address here.
The absolute majority of young men and women who identify as gay when they are teenagers (always subject to change later in life) are not subject to hostility as the findings of this poll suggest.
Your statement makes no sense at all.
You've spent much of your time trying to dilute the observed incidence of bullying experienced by LGBT kids by saying that only LGBT kids experience it, and scaling it against the entire school population (who wouldn't be the target of homophobic bullying anyway).
Now you are trying to square the circle by saying the absolute majority of LGBT kids are not subject to hostility when the data on homophobic remarks, even assuming the most improbable maximum MoE says:
...the 85 percent of gay students who regularly heard homophobic remarks could be as low as 78% or as high as 92%...
I'm sure we're now going to split the hair on what "hostility" means
, but even at it's most improbable low, 78% of lesbian, gay, bi, and trans kids regularly hear homophobic remarks
That number sounds like a clear and absolute majority to me.
That is not even the most important point. The point is that for the affected children the impact on them is severe and follow on studies are beginning to show the impact of this bullying is lifelong leading to extremely poor health outcomes.
The logical conclusion of your pattern of argument is that the incidence of sexual abuse and assault in the research make them equally as "meaningless" when scaled against the whole population, that the motives of the survivor advocacy organizations are equally as self-serving, and so no monies, services, policies, or laws should be moved to address sexual abuse and sexual assault, either.
I find that pattern of thinking unscientific, cynical and defeatist and choose to look to the better angels of our natures.